US20110170343A1

Movatterモバイル変換

Info

Publication number: US20110170343A1
Application number: US12/942,754
Authority: US
Inventors: Carlos Mazure; Richard Ferrant
Original assignee: Soitec SA
Current assignee: Soitec SA
Priority date: 2010-01-14
Filing date: 2010-11-09
Publication date: 2011-07-14
Also published as: SG173247A1; KR20110083470A; US8305803B2; FR2955204B1; TWI470771B; CN102130128B; EP2346077A1; KR101135826B1; CN102130128A; FR2955204A1; JP2011146685A; JP5420517B2; TW201126699A

Abstract

The invention relates to a memory cell having an FET transistor with a source, a drain and a floating body between the source and the drain, and an injector that can be controlled to inject a charge into the floating body of the FET transistor. The injector includes a bipolar transistor having an emitter, a base and a collector formed by the body of the FET transistor. Specifically, in the memory cell, the emitter of the bipolar transistor is arranged so that the source of the FET transistor serves as the base for the bipolar transistor. The invention also includes a memory array comprising a plurality of memory cells according to the first aspect of the invention, and to methods of controlling such memory cells.

Description

FIELD OF THE INVENTION

The field of the invention is that of semiconductor devices, and more particularly that of memory cells formed by a field effect transistor FET with floating body, and memory cell arrays comprising a plurality of memory cells of this type.

BACKGROUND OF THE INVENTION

FIG. 1 shows a cross-section of a conventional floating body DRAM (dynamic random access memory) memory cell. The conventional floating body DRAM cell is formed in a silicon-on-insulator (SOI) substrate comprising a thin film ofsilicon3 separated from thebase substrate1 by a buried oxide layer (BOX)2. Afloating body4, asource region5 and adrain region6 are formed in thethin film3 on top of the BOX2. A gatedielectric layer7 and agate electrode8 are sequentially deposited on top of thefloating body4. Thedrain region6 is connected to a bit line BL, thesource region5 is connected to a source line SL and thegate electrode8 is connected to a word line WL.

The floating body is electrically insulated by the BOX, the gate dielectric layer, the source region and the drain region. Because of this insulation, the floating body can store an electrical charge. In a data write operation in such a transistor, the floating body uses an impact ionization phenomenon to store charges, which modifies the threshold voltage of the transistor. In a data read operation, the quantity of the current flowing between the source and the drain of the transistor thus depends on the quantity of charges stored in the floating body.

In order to be able to perform alogic 1 state write operation (operation hereinafter called “write1”), a relatively high voltage, of around 2.5 times the nominal power supply voltage VDD that has to be applied to the gate electrode, must be applied to the bit line BL. Besides the fact that this high voltage is likely to damage the cell, it can also disrupt the operation of nearby memory cells. The generation of this high voltage also requires dedicated circuitry implementing charge pumps.

One technique for reducing the surface area occupied by such a floating body DRAM memory cell is described in the document US 2004/0108532. This document proposes creating a memory cell by associating a floating body horizontal FET transistor and a horizontal bipolar transistor suitable for injecting charges into the floating body. The low storage capacity of the floating body is thus increased. The horizontal bipolar transistor is more specifically arranged laterally to the FET transistor with the emitter (serving as injector) of the bipolar transistor formed in the substrate with a conductivity opposite to the latter, the base of the bipolar transistor being formed by the substrate and the collector of the bipolar transistor serving as floating body for the FET transistor.

With such an arrangement, the surface area of the memory cell is reduced to approximately 10F². However, in this arrangement, the emitter and the base consume surface area so that an even greater reduction of the memory cell surface area remains an objective to those skilled in the art.

Moreover, this memory cell has the drawback that it is difficult to accurately control the voltage of the substrate serving as base for the bipolar transistor. Since the injector is, moreover, shared between two adjacent memory cells, the result is a risk of disturbances between these neighbouring cells.

It will therefore be understood that the memory cell according to the document US 2004/0108532 is not totally satisfactory, and that there is still a need to remedy the abovementioned drawbacks in the prior art.

SUMMARY OF THE INVENTION

The aim of the invention is to propose an improved memory cell that has a reduced surface area.

To this end, the invention proposes, according to a first aspect, a memory cell comprising an FET transistor having a source, a drain and a floating body between the source and the drain, and also an injector that can be controlled to inject a charge into the floating body of the FET transistor, the injector comprising a bipolar transistor having an emitter, a base and a collector, the latter being formed by the floating body of the FET transistor. Specifically, the memory cell is characterized by the fact that the emitter of the bipolar transistor is arranged so that the source of the FET transistor serves as the base for the bipolar transistor.

Some preferred, but nonlimiting, aspects of the memory cell of this first aspect are as follows. The FET transistor can be horizontal with the emitter being arranged so that the emitter/source assembly forms a vertical stack. The emitter can be connected to an injection line that runs below the surface of the cell. The emitter can be integrated in the source. The emitter can be arranged in a bottom region of the source. The source can comprise a strongly doped central region and a lightly doped peripheral region surrounding the central region with the emitter being arranged within the peripheral region of the source under the central region of the source. The emitter can be arranged in a top region of the source. The source can comprise a lightly doped top region and a strongly doped bottom region arranged below the top region with the emitter being integrated in the top region.

Some further preferred, but nonlimiting, aspects of the memory cell of this first aspect are as follows. The bottom region of the source can be arranged below a buried insulating layer and linked to the top region of the source via a connecting passage extending through the insulating layer. The emitter can be arranged below the source. The emitter can be arranged below a buried insulating layer and can be linked to the source via a connecting passage extending through the insulating layer. The FET transistor can also comprise a gate electrode recessed in the floating body and insulated from the latter by a dielectric layer. The emitter can comprise a doped semiconductive material of conductivity opposite to that of the source. The cell can be produced on a semiconductor-on-insulator substrate. The FET transistor can be formed in a well produced in the top part of a bulk substrate.

According to another aspect, the invention relates to a memory cell comprising a FET transistor having a source, a drain and a floating body between the source and the drain, and an injector that can be controlled to inject a charge into the floating body of the FET transistor, the injector comprising a bipolar transistor having an emitter, a base and a collector formed by the floating body of the FET transistor. Specifically, this memory cell being characterized in that the emitter is connected to an injection line that runs below the surface of the cell.

According to another aspect, the invention relates to a memory array comprising a plurality of memory cells according to one or more aspects of the invention. The memory array can notably comprise a source line coupled to the source of each memory cell along a column of the array and an injection line parallel to the source line coupled to the emitter of each memory cell along said column.

According to another aspect, the invention relates to a method of programming a memory cell comprising a horizontal FET transistor having a floating body, and an injector that can be controlled to inject a charge into the floating body of the FET transistor. Specifically, alogic 1 state is programmed by applying a nominal write voltage to the gate of the FET transistor, by applying a voltage less than or equal to the nominal write voltage to the drain of the FET transistor and by applying a positive voltage to the injector.

In a cell read operation, it is also possible to apply a fraction of the nominal write voltage to the gate of the FET transistor.

A first specific embodiment relates to a memory cell comprising a horizontal FET transistor having a source connected to a source line, a drain connected to a bit line and a floating body between the source and the drain; and an injector that can be controlled to inject a charge into the floating body of the FET transistor, the injector comprising a bipolar transistor having an emitter connected to an injection line, a base and a collector formed by the floating body of the FET transistor. In this cell, the emitter of the bipolar transistor is advantageously arranged so that the emitter/source assembly forms a vertical stack, the source of the FET transistor serving as the base for the bipolar transistor.

In this memory cell, the injection line typically runs below the surface of the memory cell with the emitter integrated in the source, arranged in a bottom region thereof. The source preferably comprises a strongly doped central region and a lightly doped peripheral region surrounding the central region, and the emitter is arranged within the peripheral region of the source under the central region of the source, and preferably in a top region of the source. Alternatively, the source comprises a lightly doped top region and a strongly doped bottom region arranged below the top region, and in which the emitter is integrated in the top region. In this embodiment, the bottom region of the source can be arranged below a buried insulating layer and be linked to the top region of the source via a connecting passage extending through the insulating layer.

The emitter is preferably arranged below the source. In particular, the emitter is arranged below a buried insulating layer and is linked to the source via a connecting passage extending through the insulating layer.

In the memory cell of the invention, the FET transistor may comprise a gate electrode recessed in the floating body and insulated from the latter by a dielectric layer. Advantageously, the FET transistor may be formed in a well produced in the top part of a bulk substrate. Also, the emitter may comprise a doped semiconductive material of conductivity opposite to that of the source. The memory cell can also include a buried insulating layer.

Another embodiment of the invention relates to a memory cell comprising a FET transistor having a source, a drain and a floating body between the source and the drain; and an injector that can be controlled to inject a charge into the floating body of the FET transistor, with the injector comprising a bipolar transistor having an emitter, a base and a collector formed by the floating body of the FET transistor. The emitter is typically connected to an injection line that runs below the surface of the memory cell.

The invention also relates to a memory array comprising a plurality of memory cells as disclosed herein. In one aspect, the array further comprises a source line coupled to the source of each memory cell along a column of the array and an injection line parallel to the source line coupled to the emitter of each memory cell along said column.

Various methods of controlling a memory cell are described herein. The memory cell can be controlled to program alogic 1 state by applying a positive voltage with a magnitude less than or equal to about the nominal power supply voltage to the drain of the FET transistor; applying a positive voltage to the injector; and applying about a nominal power supply voltage to the gate of the FET transistor. Alternatively, the memory cell can be controlled to program a logic 0 state by applying a negative voltage with a magnitude less than or equal to about the nominal power supply voltage to the drain of the FET transistor; and applying about a nominal power supply voltage to the gate of the FET transistor.

Further aspects and details and alternate combinations of the elements of this invention will be apparent from the following detailed description and are also within the scope of the inventor's invention.

BRIEF DESCRIPTION OF THE DRAWINGS

Other aspects, aims and advantages of the present invention will become more apparent from reading the following detailed description of preferred embodiments of the latter, given by way of nonlimiting examples and with reference to the appended drawings in which:

FIG. 1 represents a conventional floating body DRAM cell;

FIG. 2arepresents a cross-sectional view of a memory cell according to one possible embodiment of the first aspect of the invention;

FIG. 2brepresents a possible topology of a memory array employing memory cells ofFIG. 2a;

FIG. 2crepresents a circuit diagram equivalent to the memory cell ofFIG. 2a;

FIG. 3arepresents a possible embodiment according to the invention of a memory cell using a partially depleted FET transistor on a semiconductor-on-insulator substrate;

FIG. 3brepresents a possible embodiment according to the invention of a memory cell using an FET transistor on a bulk substrate;

FIG. 3crepresents another possible embodiment of a memory cell according to the invention in which the injection line is buried below the insulating layer;

FIG. 3drepresents another possible embodiment of a memory cell according to the invention in which the source line is buried below the insulating layer;

FIG. 3erepresents a possible topology of a memory array employing memory cells ofFIG. 3d;

FIGS. 4aand4brespectively represent programming a logic 0 state in a conventional DRAM cell and in a DRAM cell according to the invention;

FIGS. 5aand5brespectively represent programming alogic 1 state in a conventional DRAM cell and in a DRAM cell according to the invention;

FIGS. 6aand6brespectively represent reading the logic state stored by a conventional DRAM cell and by a DRAM cell according to the invention;

FIGS. 7aand7brespectively represent holding the logic state stored in a conventional DRAM cell and in a DRAM cell according to the invention; and

FIG. 8 represents another possible topology of a memory array employing memory cells according to the invention.

DETAIL DESCRIPTION OF THE PREFERRED EMBODIMENTS

The preferred embodiments of the invention to be described do not limit the scope of the invention, since these embodiments are illustrations of several preferred aspects of the invention. Any equivalent embodiments are intended to be within the scope of this invention. Indeed, various modifications of the invention in addition to those shown and described herein, such as alternate useful combinations of the elements described, will become apparent to those skilled in the art from the subsequent description, and such modifications are also intended to fall within the scope of the appended claims.

Referring now toFIG. 2a, a cross-sectional view of a memory cell according to a preferred embodiment of the first aspect of the invention is represented. This cell comprises an FET transistor having a source S, a drain D and a floating body FB between the source and the drain, and an injector that can be controlled to inject a charge into the floating body FB of the FET transistor.

The memory cell is, in this case, produced on a semiconductor-on-insulator SeOI substrate, preferentially a silicon-on-insulator SOI substrate. The drain D and the source S are preferentially in contact with the buried oxide layer BOX so that the FET transistor is fully depleted. The source S can thus be shared between two adjacent memory cells (along a line of a memory array extending in the plane ofFIG. 2a; the columns of the memory array extending perpendicularly to the plane ofFIG. 2a). Such sharing makes it possible to reduce the surface area occupied by a memory cell.

As can be seen inFIGS. 3aand3b, the invention is not, however, limited to a fully depleted memory cell, but also extends to a memory cell on partially depleted SeOI (FIG. 3a) and to a memory cell produced on a bulk substrate (FIG. 3b). In the case of these figures, and in a manner conventionally known, it is necessary to isolate the cells along a line of the memory array in order to create the floating body effect. In the case ofFIG. 3a, this is conventionally done by lateral isolating trenches13 (according to the shallow trench isolation STI technique) extending depthwise from the surface of the substrate to the BOX.

In the case ofFIG. 3b, this is conventionally done bylateral isolating trenches14 extending depthwise from the surface of the substrate, and by isolating the memory cell in a well15 produced in the top part of the bulk substrate and delimited by thetrenches14. The FET transistor is situated in thesemiconductive well15, the conductivity of which is identical to that of the channel, which makes it possible to isolate it from the substrate. The well can also be arranged on alayer16 of the substrate having an opposite conductivity.

Returning to the description ofFIG. 2a, the injector consists of a bipolar transistor having anemitter15, a base and a collector formed by the floating body FB of the FET transistor. In the context of the invention, theemitter15 of the bipolar transistor is arranged so that the source S of the FET transistor serves as the base for the bipolar transistor. It will be noted that controlling the voltage of the source serving as the base for the bipolar transistor is easier than controlling the voltage of the substrate serving as the base for the bipolar transistor of the memory cell presented in the document US 2004/0108532.

The FET transistor is a horizontal transistor and the emitter of the bipolar transistor is more specifically arranged facing the source of the FET transistor so that the emitter/source assembly forms a vertical stack. In a preferred embodiment, the emitter is connected to an injection line that runs below the surface of the memory cell, thereby not consuming surface area.

Theemitter15 of the bipolar transistor can notably be integrated in the source of the FET transistor. According to a first variant embodiment represented inFIG. 2a, theemitter15 is integrated in the

source

21,22 in a bottom region thereof. According to a second variant embodiment represented inFIG. 3d, theemitter35 is integrated in the

source

31,32 in a top region thereof.

Still referring toFIG. 2a, the source conventionally comprises in a manner known per se a strongly doped central region21 (for example N+) and a lightly doped peripheral region22 (for example N−) surrounding the central region. The main purpose of the strongly dopedcentral region21 is to ensure the contact with the source line SL while the lightly dopedperipheral region22 mainly contributes to the operation of the FET transistor. Thus, it is the lightly dopedregion22 which is used in the context of the invention as the base for the bipolar transistor.

In the context of this variant, theemitter15 is arranged so as to be positioned below thecentral region21 of the source while being in contact with theperipheral region22 of the source and isolated from the floating body FB by the peripheral22 of the source. The emitter is, in this case, totally integrated in the source electrode between the central21 and peripheral22 regions thereof.

In the case of a memory cell made of fully depleted SeOI (seeFIG. 2a), and unlike the case of a memory cell made of partially depleted SeOI (FIG. 3a) or a memory cell on bulk substrate (FIG. 3b), the BOX can help to insulate theemitter15 from the floating body. It will be noted that the drain D can also have a strongly doped central region11 (N+ in the example) and a lightly doped peripheral region12 (N− in the example) surrounding the central region.

In a manner known conventionally, the drain D is linked to a bit line BL. This bit line BL can extend along a line of the memory array, coming into contact with the drain of each of the memory cells arranged along this line.

As can be seen inFIG. 2b, the source S is connected to a source line SL. This source line SL typically extends perpendicularly to the bit line BL, coming into contact with the source (typically the strongly doped region of the source) of each of the memory cells arranged along a column of the memory array.

In the case ofFIGS. 2aand2b, the source line SL therefore runs immediately below the surface of the SeOI substrate. Moreover, in the case of theseFIGS. 2aand2bin which the source is shared between two adjacent cells, a single source line SL can be used to address the source of the two memory cells. Theemitter15 is linked to an injection line IL which extends parallel to the source line SL so as to address the memory cells arranged along a column of the memory array.

It will be understood from the foregoing that, in the case ofFIG. 2a, the injection line IL is arranged below the source line SL. The result of this is a particularly compact arrangement inasmuch as the injection line IL does not consume any surface region.

FIG. 2brepresents a topology of a memory array according to the invention with the source line SL shared between two adjacent cells. This arrangement is thus even more compact. Based on the topology ofFIG. 2b, the surface area of the memory cell can thus be reduced to approximately 4 F²in the case where a standalone memory is being developed. While this cell is not totally insensitive to disturbances, nevertheless the level of disturbances remains low and in any case less than the level of disturbance encountered with conventional memory cells.

In the case of an embedded memory, given that the requirements are greater for the transistors (need to provide spacers to avoid cross-excitation phenomena), the surface area occupied by the transistors is greater. A memory cell surface area of the order of 15 F²to 18 F²is then achieved. It does, however, offer the benefit of being totally insensitive to the disturbances.

Another possible topology, less compact but totally insensitive to disturbances, is represented inFIG. 8. In this topology, the word line WL controls two adjacent physical rows which share a common source line SL. This particular arrangement eliminates most of the sources of mutual disturbances since the cells situated on this double word line WL are all used and it can be legitimately assumed that the write circuits are more powerful than the disturbances. On the other hand, the two cells situated one on top of the other must, imperatively, be connected to two different bit lines BL1, BL2, otherwise their information would become mixed during the WL access. This has a direct impact on the surface area of the cell since it must be wide enough to leave two lines of metal alongside one another. While the latter point is generally prohibitive for standalone memories, it is, on the other hand, perfectly legitimate in the case of embedded memories inasmuch as the surface area of the cell is around 4F*4F=16 F².

As is represented inFIG. 2a, the FET transistor also has a gate electrode positioned above the floating body, on the surface of the memory array, and insulated from the floating body by a dielectric layer. The gate electrode is linked to a word line WL which typically extends along a column of the memory array.

According to another embodiment not represented, the gate electrode is recessed in the floating body and insulated therefrom by a dielectric layer. An RCAT (recessed channel array transistor) type transistor is thus defined.

FIG. 2cis a circuit diagram equivalent to the memory cell according to the first aspect of the invention. In this figure, the FET transistor—in this case N—has thereference9 and the bipolar transistor—in this case PNP—has thereference10. As previously discussed, the source of theFET transistor9 connected to the source line SL serves as the base for thebipolar transistor10, while the body of theFET transistor9 serves as the collector for thebipolar transistor10.

Referring toFIG. 3dillustrating another variant embodiment, theemitter35 is integrated in the

source

31,32 in a top region thereof. In the context of this variant, the source S comprises a lightly dopedtop region32 and a strongly dopedbottom region31 arranged below the top region, theemitter35 being integrated in thetop region32. As represented in thisFIG. 3d, thebottom region31 of the source can in particular be arranged below a buried insulating layer in the substrate, for example the BOX layer in the case of an SeOI substrate, while being linked to thetop region32 of the source via a connectingpassage33 extending through the insulating layer. The strongly dopedbottom region31 of the source (N+ doping in the example concerned) is more specifically arranged, below the insulating layer, in aregion36 of the substrate that has an opposite conductivity (P− region in the example).

In the context of this variant, the source line SL is buried; it can notably be buried directly below the insulating layer BOX as represented inFIG. 3d. Theemitter35 of the bipolar transistor is linked to an injection line IL which extends directly below the surface of the substrate (this injection line IL is provided instead of the source line of a conventional cell) and which, because of this, is easier to access. Also in this variant, the vertical stack of the source and the emitter has the particular consequence that the injection line consumes no surface area.

FIG. 3erepresents a possible topology of a memory array employing memory cells ofFIG. 3d. In this topology, the source line SL extends parallel to the injection line IL along a column of the memory array, while being buried below the injection line, under the BOX layer. In this variant embodiment, the injection line IL is simpler to access.

FIG. 3crepresents another possible embodiment, in which the vertical stack of the source and the emitter is obtained by arranging theemitter25 below the

source

21,22. It will be noted that this embodiment corresponds to the variant ofFIG. 3dwith the emitter and the strongly doped region of the source simply interchanged. In practice, in this other embodiment, the emitter25 (P+) is buried under the insulating layer, in aregion26 of the substrate that has an opposite conductivity (N− in the example) and is linked to theperipheral region22 of the source via a connectingpassage23 extending through the insulating layer.

The source line SL conventionally extends immediately below the surface of the substrate, while the injection line IL extends under the BOX layer parallel to the source line SL. Thus, the injection line IL does not consume any surface area.

The injection line is preferentially produced using a doped semiconductive material. This is a P-type doping (notably, a P+ doping) of the material of the substrate in the examples represented in the figures in which the FET transistor is of the N type (P-type floating body). It will be understood that the invention is not limited to an N FET transistor, but also extends to the case of a P FET transistor (N-type floating body).

By using the techniques conventionally used in CMOS technology to produce such a doping of the substrate in order to produce the injection line IL, the latter does not need to be metallized so that metal wiring can be avoided.

FIGS. 4a-4b,5a-5b,6a-6band7a-7bshow the equivalent circuit diagram ofFIG. 2ain order to compare the different programming, read and hold operations for a conventional memory cell of the type of that illustrated inFIG. 1 (FIGS. 4a,5a,6aand7a) and for a memory cell according to the invention (FIGS. 4b,5b,6band7b).

In these various figures:

FIGS. 4aand4billustrate the WRITE 0 operation for programming the logic 0 state,

FIGS. 5aand5billustrate theWRITE 1 operation for programming thelogic 1 state,

FIGS. 6aand6billustrate the READ operation for reading the logic state,

FIGS. 7aand7billustrate the HOLD operation for holding the logic state.

The tables below illustrate examples of voltages that have to be applied to the various lines to control the cell and carry out these various operations.

TABLE 1

Memory cell of FIG. 1 (prior art)

abbrev.	name of connection	connected to

BL	bit line	drain of FET
WL	word line	gate of FET
SL	source line	source of FET
IL	injection line	emitter of bipolar
	body of FET	collector of bipolar
	source of FET	base of bipolar

nominal voltage - VDD supply voltage

BL	WL	SL
drain of FET	gate of FET	source ofFET

WRITE

1	2.5	VDD	VDD	0 V
	WRITE 0	−0.25	VDD	VDD	0 V
	READ	0.25	VDD	VDD	0 V
	HOLD	0	V	0 V	0 V

TABLE 2

Memory cell according to the invention

			IL
BL	WL	SL	emitter
drain of FET	gate of FET	source of FET	ofbipolar

WRITE

1	0.75	VDD	VDD	0 V	0.75	VDD
WRITE 0	−0.25	VDD	VDD	0 V	0	V
READ	0.25	VDD	0.5 VDD	0 V	0	V
HOLD	0	V	0 V	0 V	0	V

It will first of all be observed that the voltage that has to be applied to the word line WL to carry out the read operation READ is, moreover, lowered to a fraction of the nominal power supply voltage VDD, for example about 0.5 VDD. This makes it possible to lower the overall consumption level of the cell. Furthermore, inasmuch as this voltage is situated between the two threshold voltage levels (“0” and “1”), the read selection leads to the presence of a current for the lowest threshold voltage (in principle, “0” by convention), and the absence of current in the other case. The design of the detection amplifier is then reduced to a “simple” comparator that is much easier to design than if there are two currents of different amplitudes that must be compared to references, references that must be known and that must be trusted.

Moreover, as already described in the introduction, a voltage of around 2.5 times the nominal write voltage VDD must be applied to the drain of the conventional memory cell to carry out theWRITE 1 operation.

By using the injection of charge into the floating body by the injector, the invention makes it possible to lower the voltage needed to carry out thisWRITE 1 operation. In particular, the voltage that has to be applied to the drain of a memory cell according to the invention is less than or equal to the nominal voltage VDD, for example about 0.75 VDD as indicated in Table 2 above

The fact that all the programming operations are carried out with low voltages applied to all thepower supply lines 1 offers a number of advantages. First of all, the power consumption is reduced and the design of the peripheral circuits is simplified. In particular, the circuitry needed to generate high voltages (2.5 VDD for the conventional memory cell, notably for theWRITE 1 operation) is not necessary.

Moreover, by reducing the voltage, damage to the memory cell is avoided. Disturbance of the neighbouring cells is also avoided. Thus, a disturbance potentially generated via the bit line BL in a write operation on a neighbouring cell corresponds to a voltage of 0.75 VDD (WRITE 1) or −0.25 VDD (WRITE 0) on the bit line BL in the example in question (with WL=SL=IL=0 V). These relatively low voltages are not likely to generate significant disturbances.

A disturbance potentially generated via the word line WL in aWRITE 1 write operation on a neighbouring cell corresponds to respective voltages of VDD on the word line WL and of 0.75 VDD on the injection line IL (and with BL=SL=0V). An injection of charge (holes in the example described here) into the floating body via the bipolar transistor is then likely to occur.

In a WRITE 0 write operation on a neighbouring cell, WL is at VDD while BL=SL=IL=0 V. The bipolar transistor (PNP type in the example described here) is then OFF and all the holes are likely to be evacuated via at negative bit line.

In order to simply overcome these disturbance phenomena, a choice can be made to not remain passive by writing to the stressed cells given that they are located on the active word line WL (which is tantamount to carrying out valid WRITE operations on all the cells along WL by applying the appropriate voltages to all the corresponding bit lines). This conditions the partitioning of the memory (requires all the cells along the word line WL to belong to the same word) without, however, this posing any problem given that all the stressed cells can be used directly by the application. In practice, if the word is wider, the information bit rate is greater and ultimately the consumption is lower.

In general and unless specifically stated to the contrary elsewhere, the following approximate voltages and voltage ranges are preferred. Preferred READ voltages on BL can be any voltage in the range about 0.1-0.4 VDD, ex., about 0.1 VDD, or about 0.2 VDD, or about 0.25 VDD, or about 0.3 VDD, or about 0.4 VDD. Preferred READ voltages on WL can be any voltage in the range about 0.3-0.7 VDD, ex., about 0.3 VDD, or about 0.4 VDD, or about 0.5 VDD, or about 0.6 VDD, or about 0.7 VDD.Preferred WRITE 1 voltages on BL and IL can be any voltage in the range about 0.6-0.9 VDD, ex., 0.6 VDD, or about 0.7 VDD, or about 0.75 VDD, or about 0.8 VDD, or about 0.9 VDD. Preferred WRITE 0 voltages on BL can be any voltage in the range about −0.4 to −0.1 VDD, ex., about −0.4 VDD, or about −0.3 VDD, or about −0.25 VDD, or about −0.2 VDD, or about −0.1 VDD.Preferred WRITE 1 or WRITE 0 voltages on WL can be about 1.0 VDD (alternatively, preferred read voltages on BL being any voltage in the range about 0.8-1.0 VDD, ex., about 0.8 VDD, or about 0.9 VDD, or about 1.0 VDD. Further, in cases where a preferred voltage is 0.0 VDD, preferred voltages can be any voltage in the range about −0.2-0.2 VDD, ex., about −0.2 VDD, or about −0.1 VDD, or about 0.0 VDD, or about 0.1 VDD, or about 0.2 VDD.

It will have been understood from the foregoing that the invention is not limited to a memory cell according to its first aspect, but also extends to a memory array comprising a plurality of memory cells according to the first aspect of the invention, and to a method of controlling such a memory cell.

The words “significant” and “likely” (and similar words of degree) are used here to mean within acceptable and expected limits for the intended use, usually but not exclusively commercial uses. For example, in the phrase “significant disturbances”, disturbances to memory cell operation are considered significant if, ex., they are greater the level of disturbance encountered with conventional, commercially-acceptable memory cells in conventional applications. Similarly the phrase “not likely to generate significant disturbances” is considered to mean that such “significant disturbances” occur less than 25%, or less than 5%, or less than 1%, or less than 0.2%, or less than 0.1%, if the time. It should be understood that this invention is not limited to commercial uses; intended uses include research uses, special purpose uses, and so forth, and that “significant” is judged in the context of these other uses.